Improve FencedCodeBlockSyntax

lib/src/block_syntaxes/fenced_code_block_syntax.dart

@@ -4,7 +4,6 @@
 
 import '../ast.dart';
 import '../block_parser.dart';
-import '../charcode.dart';
 import '../patterns.dart';
 import '../util.dart';
 import 'block_syntax.dart';
@@ -19,77 +18,118 @@ class FencedCodeBlockSyntax extends BlockSyntax {
   const FencedCodeBlockSyntax();
 
   @override
-  bool canParse(BlockParser parser) {
-    final match = pattern.firstMatch(parser.current);
-    if (match == null) return false;
-    final codeFence = match.group(1)!;
-    final infoString = match.group(2);
-    // From the CommonMark spec:
-    //
-    // > If the info string comes after a backtick fence, it may not contain
-    // > any backtick characters.
-    return codeFence.codeUnitAt(0) != $backquote ||
-        !infoString!.codeUnits.contains($backquote);
+  Node parse(BlockParser parser) {
+    final openingFence = _FenceMatch.fromMatch(pattern.firstMatch(
+      escapePunctuation(parser.current),
+    )!);
+
+    var text = parseChildLines(
+      parser,
+      openingFence.marker,
+      openingFence.indent,
+    ).join('\n');
+
+    if (parser.document.encodeHtml) {
+      text = escapeHtml(text);
+    }
+    if (text.isNotEmpty) {
+      text = '$text\n';
+    }
+
+    final code = Element.text('code', text);
+    if (openingFence.hasLanguage) {
+      var language = decodeHtmlCharacters(openingFence.language);
+      if (parser.document.encodeHtml) {
+        language = escapeHtmlAttribute(language);
+      }
+      code.attributes['class'] = 'language-$language';
+    }
+
+    return Element('pre', [code]);
   }
 
-  @override
-  List<String> parseChildLines(BlockParser parser, [String? endBlock]) {
-    endBlock ??= '';
+  String _removeIndentation(String content, int length) {
+    final text = content.replaceFirst(RegExp('^\\s{0,$length}'), '');
+    return content.substring(content.length - text.length);
+  }
 
+  @override
+  List<String> parseChildLines(
+    BlockParser parser, [
+    String openingMarker = '',
+    int indent = 0,
+  ]) {
     final childLines = <String>[];
+
     parser.advance();
 
+    _FenceMatch? closingFence;
     while (!parser.isDone) {
       final match = pattern.firstMatch(parser.current);
-      if (match == null || !match[1]!.startsWith(endBlock)) {
-        childLines.add(parser.current);
+      closingFence = match == null ? null : _FenceMatch.fromMatch(match);
+
+      // Closing code fences cannot have info strings:
+      // https://spec.commonmark.org/0.30/#example-147
+      if (closingFence == null ||
+          !closingFence.marker.startsWith(openingMarker) ||
+          closingFence.hasInfo) {
+        childLines.add(_removeIndentation(parser.current, indent));
         parser.advance();
       } else {
         parser.advance();
         break;
       }
     }
 
+    // https://spec.commonmark.org/0.30/#example-127
+    // https://spec.commonmark.org/0.30/#example-128
+    if (closingFence == null && childLines.last.trim().isEmpty) {
+      childLines.removeLast();
+    }
+
     return childLines;
   }
+}
 
-  @override
-  Node parse(BlockParser parser) {
-    // Get the syntax identifier, if there is one.
-    final match = pattern.firstMatch(parser.current)!;
-    final endBlock = match.group(1);
-    var infoString = match.group(2)!;
-
-    final childLines = parseChildLines(parser, endBlock);
+class _FenceMatch {
+  _FenceMatch._({
+    required this.indent,
+    required this.marker,
+    required this.info,
+  });
 
-    // The Markdown tests expect a trailing newline.
-    childLines.add('');
+  factory _FenceMatch.fromMatch(RegExpMatch match) {
+    String marker;
+    String info;
 
-    var text = childLines.join('\n');
-    if (parser.document.encodeHtml) {
-      text = escapeHtml(text);
+    if (match.namedGroup('backtick') != null) {
+      marker = match.namedGroup('backtick')!;
+      info = match.namedGroup('backtickInfo')!;
+    } else {
+      marker = match.namedGroup('tilde')!;
+      info = match.namedGroup('tildeInfo')!;
     }
-    final code = Element.text('code', text);
 
-    // the info-string should be trimmed
-    // http://spec.commonmark.org/0.22/#example-100
-    infoString = infoString.trim();
-    if (infoString.isNotEmpty) {
-      // only use the first word in the syntax
-      // http://spec.commonmark.org/0.22/#example-100
-      final firstSpace = infoString.indexOf(' ');
-      if (firstSpace >= 0) {
-        infoString = infoString.substring(0, firstSpace);
-      }
-      infoString = decodeHtmlCharacters(infoString);
-      if (parser.document.encodeHtml) {
-        infoString = escapeHtmlAttribute(infoString);
-      }
-      code.attributes['class'] = 'language-$infoString';
-    }
+    return _FenceMatch._(
+      indent: match[1]!.length,
+      marker: marker,
+      info: info.trim(),
+    );
+  }
 
-    final element = Element('pre', [code]);
+  final int indent;
+  final String marker;
 
-    return element;
-  }
+  // The info-string should be trimmed,
+  // https://spec.commonmark.org/0.30/#info-string.
+  final String info;
+
+  // The first word of the info string is typically used to specify the language
+  // of the code sample,
+  // https://spec.commonmark.org/0.30/#example-143.
+  String get language => info.split(' ').first;
+
+  bool get hasInfo => info.isNotEmpty;
+
+  bool get hasLanguage => language.isNotEmpty;
 }

lib/src/patterns.dart

@@ -21,7 +21,9 @@ final blockquotePattern = RegExp(r'^[ ]{0,3}>[ ]?(.*)$');
 final indentPattern = RegExp(r'^(?:    | {0,3}\t)(.*)$');
 
 /// Fenced code block.
-final codeFencePattern = RegExp(r'^[ ]{0,3}(`{3,}|~{3,})(.*)$');
+final codeFencePattern = RegExp(
+  r'^([ ]{0,3})(?:(?<backtick>`{3,})(?<backtickInfo>[^`]*)|(?<tilde>~{3,})(?<tildeInfo>.*))$',
+);
 
 /// Fenced blockquotes.
 final blockquoteFencePattern = RegExp(r'^>{3}\s*$');
@@ -199,6 +201,10 @@ final htmlBlockPattern = RegExp(
     '(?<condition_7>(?:$namedTagDefinition)\\s*\$))',
     caseSensitive: false);
 
+/// ASCII punctuation characters.
+// see https://spec.commonmark.org/0.30/#unicode-whitespace-character.
+const asciiPunctuationCharacters = r'''!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~''';
+
 /// A pattern to match HTML entity references and numeric character references.
 // https://spec.commonmark.org/0.30/#entity-and-numeric-character-references
 final htmlCharactersPattern = RegExp(

lib/src/util.dart

@@ -147,3 +147,20 @@ extension MatchExtensions on Match {
   /// Returns the whole match String
   String get match => this[0]!;
 }
+
+/// Escapes the ASCII punctuation characters after backslash(`\`).
+String escapePunctuation(String input) {
+  final buffer = StringBuffer();
+
+  for (var i = 0; i < input.length; i++) {
+    if (input.codeUnitAt(i) == $backslash) {
+      final next = i + 1 < input.length ? input[i + 1] : null;
+      if (next != null && asciiPunctuationCharacters.contains(next)) {
+        i++;
+      }
+    }
+    buffer.write(input[i]);
+  }
+
+  return buffer.toString();
+}

test/common_mark/backslash_escapes.unit

@@ -75,5 +75,5 @@ bar</p>
 foo
 ```
 <<<
-<pre><code class="language-foo\+bar">foo
+<pre><code class="language-foo+bar">foo
 </code></pre>

test/common_mark/block_quotes.unit

@@ -96,8 +96,7 @@ foo
 <pre><code>foo
 </code></pre>
 </blockquote>
-<pre><code>
-</code></pre>
+<pre><code></code></pre>
 >>> Block quotes - 238
 > foo
     - bar

test/common_mark/fenced_code_blocks.unit

@@ -61,8 +61,7 @@ aaa
 >>> Fenced code blocks - 126
 ```
 <<<
-<pre><code>
-</code></pre>
+<pre><code></code></pre>
 >>> Fenced code blocks - 127
 `````
 
@@ -72,7 +71,6 @@ aaa
 <pre><code>
 ```
 aaa
-
 </code></pre>
 >>> Fenced code blocks - 128
 > ```
@@ -105,7 +103,7 @@ bbb
 aaa
 ```
 <<<
-<pre><code> aaa
+<pre><code>aaa
 aaa
 </code></pre>
 >>> Fenced code blocks - 132
@@ -116,7 +114,7 @@ aaa
   ```
 <<<
 <pre><code>aaa
-  aaa
+aaa
 aaa
 </code></pre>
 >>> Fenced code blocks - 133
@@ -126,9 +124,9 @@ aaa
   aaa
    ```
 <<<
-<pre><code>   aaa
-    aaa
-  aaa
+<pre><code>aaa
+ aaa
+aaa
 </code></pre>
 >>> Fenced code blocks - 134
     ```
@@ -160,7 +158,6 @@ aaa
 <<<
 <pre><code>aaa
     ```
-
 </code></pre>
 >>> Fenced code blocks - 138
 ``` ```
@@ -175,7 +172,6 @@ aaa
 <<<
 <pre><code>aaa
 ~~~ ~~
-
 </code></pre>
 >>> Fenced code blocks - 140
 foo
@@ -245,6 +241,5 @@ foo
 ``` aaa
 ```
 <<<
-<pre><code></code></pre>
-<pre><code>
+<pre><code>``` aaa
 </code></pre>

test/common_mark/lists.unit

@@ -279,7 +279,6 @@ The number of windows in my house is
 </li>
 </ul>
 <pre><code>- c
-
 </code></pre>
 >>> Lists - 319
 - a

test/gfm/backslash_escapes.unit

@@ -75,5 +75,5 @@ bar</p>
 foo
 ```
 <<<
-<pre><code class="language-foo\+bar">foo
+<pre><code class="language-foo+bar">foo
 </code></pre>

test/gfm/block_quotes.unit

@@ -96,8 +96,7 @@ foo
 <pre><code>foo
 </code></pre>
 </blockquote>
-<pre><code>
-</code></pre>
+<pre><code></code></pre>
 >>> Block quotes - 216
 > foo
     - bar